738
Bioconjugate Chem. 2004, 15, 738−746
Amphiphilic Corroles Bind Tightly to Human Serum Albumin Atif Mahammed,† Harry B. Gray,*,‡ Jeremy J. Weaver,‡ Karn Sorasaenee,‡ and Zeev Gross*,† Department of Chemistry and Institute of Catalysis Science and Technology, Technion - Israel Institute of Technology, Haifa 32,000, Israel, and Department of Chemistry and Beckman Institute, California Institute of Technology, Pasadena, California 91125. Received September 30, 2003; Revised Manuscript Received May 3, 2004
Amphiphilic 2,17-bis-sulfonato-5,10,15(trispentafluorophenyl)corrole (2) and its Ga and Mn complexes (2-Ga and 2-Mn) form tightly bound noncovalent conjugates with human serum albumin (HSA). Protein-induced changes in the electronic absorption, emission, and circular dichroism spectra of these corroles, as well as results obtained from HPLC profiles of the conjugates and selective fluorescence quenching of the single HSA tryptophan, are interpreted in terms of multiple corrole:HSA binding sites. High-affinity binding sites, close to the unique tryptophan, are fully occupied at very low concentrations. At biologically relevant HSA concentrations (2-3 orders of magnitude larger than those employed in our studies), all corroles (2, 2-Ga, and 2-Mn) may be considered as fully conjugated.
INTRODUCTION
The bioinorganic chemistry of porphyrins is a broad field, ranging from natural to completely synthetic systems. The field also includes work on a large variety of porphyrin-like molecules (porphyrinoids), such as the chlorophyll-related chlorins and bactereochlorins as well as completely synthetic phthalocyanines and expanded porphyrins (1, 2). A great deal of research has been devoted to the interactions of porphyrinoids with biomolecules, owing to their tendency to accumulate in cancer cells and to intercalate DNA (3, 4). Surprisingly, although corroles have been known for almost four decades (5, 6), there has been very little work on their interactions with biological molecules. This situation does not reflect a lack of interest in these tetrapyrrolic macrocycles (they are related to the cobalt-chelating corrin in vitamin B12 by virtue of an identical carbon skeleton and to porphyrins by being fully conjugated), but only because their syntheses heretofore have been long and tedious. This situation has recently changed in dramatic fashion: facile synthetic methodologies have been introduced during the last 5 years, allowing for the preparation of more than 100 new corroles (7-17); transition metal complexes (Cr, Mn, Fe, Rh) of the most extensively investigated derivative (tris(pentafluorophenyl)corrole, 1) function as versatile catalysts for a variety of organic reactions (1822); and the GaIII and AlIII complexes of corrole 1 are highly fluorescent, with quantum yields (0.37 up to 0.76) that exceed those of all other porphyrinoids (23, 24). In addition, it has been discovered recently that electrophilic substitution of 1 offers a facile and highly selective synthetic method for the preparation of corroles with polar headgroups located in only one-half of the macrocycle (amphiphilic corroles) (25, 26). This finding is key for the utilization of corroles in biological systems, since amphiphilic porphyrinoids have been shown to be the molecules of choice for many applications (27-29). As the most abundant protein in blood plasma, human serum albumin (HSA) plays a role in many biological processes (30-32). Available are high-resolution X-ray † ‡
Technion - Israel Institute of Technology. California Institute of Technology.
structures of HSA as well as information about its binding affinities to many natural and synthetic molecules (33). However, with the exception of one preliminary report, there are no X-ray structures of HSA conjugates with any porphyrinoid or related macrocyclic molecule (34). Accordingly, information about noncovalent binding of porphyrinoids to HSA comes from data acquired by spectroscopic methods (35). Selected results from such investigations are set out in Table 1 (36-45). In many cases, the conclusions about the number of binding sites and the association constants are based on a limited number of spectroscopic measurements, a likely reason for the significant scattering in published results that is evident from inspection of rows 6-9. Nevertheless, it is clear that only negatively charged porphyrinoids associate with HSA, and, in most cases, the protein has one well-defined high-affinity site and several other lower-affinity sites (42). Another important finding is the inverse relationship between the association constants and the number of charged groups. The strongest association to HSA occurs with porphyrinoids that contain only two carboxylato or sulfonato groups located on the pyrrole, benzopyrrole, or phenyl rings. Intriguingly, these dipolar and amphiphilic molecules are also the most active derivatives in photodynamic therapy. A major drawback, however, is that porphyrins or phthalocyanines with a limited (less than four) number of polar headgroups are not readily accessible, as illustrated by the bis-sulfonated phthalocyanines that are obtained and used as a mixture of at least eight nonseparable isomers, even after separation of derivatives with a different degree of sulfonation (46). In sharp contrast to all other synthetic porphyrinoids, the amphiphilic 2,17-bis-sulfonato-5,10,15(trispentafluorophenyl)corrole (2) with its two precisely located sulfonate groups can be obtained in very simple synthetic steps from readily available starting materials (Scheme 1) (25, 26). Accordingly, we decided to investigate the association of 2 and its gallium(III) and manganese(III) complexes (2-Ga, and 2-Mn, respectively) to HSA as a first step toward utilization of corroles in bioinorganic applications. Employing several complementary physical methods, we have found that all three corroles form noncovalent conjugates with HSA. The binding affinities of these conjugates decrease ac-
10.1021/bc034179p CCC: $27.50 © 2004 American Chemical Society Published on Web 06/19/2004
Amphiphilic Corroles
Bioconjugate Chem., Vol. 15, No. 4, 2004 739
Table 1. Association Constants of HSA:Porphyrinoid Conjugatesa no. 1 2 3 4
ligand uroporphyrin I uroporphyrin I heptacarboxylporphyrin coproporphyrin I
5 6 7 8 9 10
coproporphyrin I protoporphyrin IX protoporphyrin IX protoporphyrin IX protoporphyrin IX hemin
11 12 13
mesoporphyrin IX Mg-mesoporphyrin IX Al (PcS1)
14
Al (PcS2) (opp)
15
Al (PcS2) (adj)
16
Al (PcS3)
17 18 19
Al(PcS4) Al(PcS2) tpss
20
Zn(2,6-Cl, 3SO3H)tpp
no. of binding sites N)1 N)1 N)1 N)1 another weak site N)1 N)1 N)1 N)1 N)1 Nstrong ) 1 Nweak g 4 Nstrong ) 1 Nweak ) 8 Nstrong ) 1 Nweak ) 8 Nstrong cooperative site ) 1 Nweak ) 8 Nstrong cooperative site ) 1 Nweak ) 8 nonsaturable
N)7
association constant, K (M-1)
method
ref
(8.8 ( 0.51) × 2.0 × 104 (2.39 ( 0.16) × 105 (1.61 ( 0.11) × 106 (3.1 ( 0.3) × 105 (1.3 ( 0.11) × 105 (9.34 ( 0.3) × 106 0.5 × 106 1.3 × 106 2.8 × 108 5 × 107 (2.5 ( 0.7) × 107 (1.7 ( 0.5) × 107 (4.2 ( 0.8) × 107 (3.8 ( 0.9) × 104 (2.8 ( 1.1) × 107 (4.0 ( 0.7) × 104 (5.8 ( 0.6) × 106 (2.7 ( 0.8) × 104 (2.5 ( 1.2) × 106 (2.5 ( 0.7) × 104 104 (5.0 ( 2) × 106 (0.2 ( 0.1) × 106 -
b c b b b d b c c e f&c
[36] [37] [36] [36]
e e f
[41] [40] [42]
f
[42]
f
[42]
f
[42]
f f f c f
[42] [43] [44]
104
[37] [36] [38] [37] [39] [40]
[45]
a
Abbreviations: PcS1, PcS2, PcS3, PcS4: phthalocyanine with 1, 2, 3, and 4 sulfonato groups, respectively, where opp and adj stand for the relative positions of the sulfonato groups in PcS2 on opposite and adjacent benzopyrrole rings, respectively; tpss: 5,10,15,20tetrakis(p-sulfonatophenyl)porphyrin; (2,6-Cl, 3-SO3H)tpp: 5,10,15,20-tetrakis(2,6-dichloro-3-sulfonatophenyl)porphyrin. b Affinity capillary electrophoresis. c Fluorescence quenching of HSA by the porphyrinoid. d Equilibrium dialysis. e HSA-initiated changes in the fluorescence of the porphyrinoid. f HSA-initiated changes in the absorbance of the porphyrinoid.
Scheme 1
cording to 2 e 2-Ga < 2-Mn. It is apparent that corrole 2 and its metal complexes will be fully bound to HSA under physiological conditions; accordingly, any biological activity observed for these molecules likely will be attributable to their protein conjugates. EXPERIMENTAL PROCEDURES
Chemicals. Preparations of corrole 2 and its metal complexes are described in previous publications (25, 26). HSA was purchased from Sigma (essentially fatty acid free, catalog no. A 1887) and used as received. Throughout these studies, 0.1 M phosphate buffer solutions of pH 7.0 were used. The extinction coefficients of the corroles under these conditions are as follows: 2: (414 nm) ) 71000 M-1 cm-1; 2-Ga: (424 nm) ) 75000 M-1 cm-1; 2-Mn: (422 nm) ) 21000 M-1 cm-1; and for HSA, (280 nm) ) 37400 M-1 cm-1. Importantly, perfectly linear plots of OD vs corrole concentration were obtained in the range of 10-6 to 10-4 M, ruling out self-aggregation phenomena at the concentrations that were used in these studies. Further support for this conclusion comes from the practically identical extinction coefficients measured in aqueous and polar organic solvents (CH3CN, MeOH). Electronic Spectroscopy. UV/Vis absorption measurements were performed on a HP8452A diode array spectrophotometer, using 3 mL of corrole solutions at 2-10 × 10-5 M and adding either microliter portions of HSA solution or milligram amounts of solid HSA.
Emission Spectroscopy. The fluorescence of HSA or the corroles was measured on an AB2 Luminescence Spectrometer. For titration of HSA’s fluorescence, microliter portions of concentrated corrole solutions were added to 3 mL solutions of HSA wherein the absorbance at the excitation wavelength of 280 nm was less than 0.15. For titration of 2 or 2-Ga, either microliter portions of a concentrated HSA solution or milligram amounts of solid HSA were added to 3 mL solutions of the corroles until no further changes in the emission intensity occurred. Circular Dichroism Spectroscopy. CD spectra were recorded in the visible range on an Aviv model 62A DS Circular Dichroism Spectrometer using 3 mL of corrole solutions at 2-10 × 10-5 M and adding either microliter portions of concentrated HSA solution or milligram amounts of solid HSA. UV-CD spectra were recorded using an HSA concentration of 1.8 × 10-6 M (in a 4-mm cuvette) and 3.7 × 10-6 M (in a 10-mm cuvette) with varying ratios of 2-Ga. Reversed Phase HPLC. HPLC analyses were performed on Merck Hitachi HPLC linear gradient using LiChrospher 100, a RP-18 column (5 µm, 4 × 125 mm), and a flow rate of 1 mL/min. The mobile phase was 100% water at time zero, turning to 100% acetonitrile within 15 min, continued for 5 min with acetonitrile only, turning back to 100% water within another 15 min, and continued for 10 min with water only. Eluted components were monitored spectrophotometrically at 436, 426, and 486 nm for 2, 2-Ga, and 2-Mn, respectively, wavelengths at which the absorbances of free and conjugated corroles are identical. An initial concentration of [HSA] ) 2.5 × 10-4 M was used for experiments at [HSA]:[corrole] ratio of 5:1 and [corrole] ) 5 × 10-5 M for the [HSA]:[corrole] ratio of 1:5 experiments. The cells were made of quartz with a 10 mm path length.
740 Bioconjugate Chem., Vol. 15, No. 4, 2004
Figure 1. Changes in the UV-vis spectra of corrole 2 (a) and its metal complexes 2-Ga (b), and 2-Mn (c), upon titration with increasing amounts of HSA. [2] ) 2.2 × 10-5 M, [2-Ga] ) 2.7 × 10-5 M, [2-Mn] ) 10-4 M; [HSA] ) 0-10-4 M. RESULTS AND DISCUSSION
Binding of Corroles to HSA. Electronic spectra of corroles when treated with increasing amounts of HSA are shown in Figure 1. For 2 and 2-Ga, respective shifts of the Soret bands from 416 to 424 nm and from 424 to 430 nm are observed. For metal-free 2, this effect could indicate changes in the protonation state of the inner N4 core from triply to doubly protonated. The pKa for this process has been determined to be 5.2 in water (47), and it might be quite different in a more hydrophobic environment. We do not believe this explanation is correct: the changes in the Q-bands upon titration with HSA are very different from those observed upon NH-deprotonation of 2 in aqueous solutions (47), so the results are likely attributable to a “solvent effect” that would manifest itself in the interior of the protein. The same interpretation also accounts for the Soret shift of 2-Ga. Spectral changes for 2-Mn were most pronounced: most bands just decreased in intensity, but the 480 nm band was clearly replaced by a 492 nm band. It is known that this particular band (No. V in manganese(III) porphyrins) is very sensitive to axial ligation. Adding imidazole to an aqueous solution of 2-Mn gave almost the same spectrum as addition of HSA, suggesting the participation of histidine residues in the binding of this metal complex. In all three cases, there were no more spectral changes after the addition of about 1 equiv of HSA, suggesting very strong association. More detailed information was obtained by following the intensity changes at selected wavelengths (Figure 2). The majority of the intensity increase at 438 and 428 nm for 2 and 2-Ga, respectively,
Mahammed et al.
Figure 2. UV-vis spectral changes at selected wavelengths as a function of the [HSA]:[corrole] ratio for (a) 2, (b) 2-Ga, and (c) 2-Mn.
is complete much before a 1:1 ratio is achieved, suggesting that there are multiple HSA binding sites for these corroles. This phenomenon is even more pronounced for 2-Mn: the intensity at 422 nm decreases sharply up to a [HSA]:[2-Mn] ratio of 0.04 and increases slightly when a 1:1 ratio is approached. Taken together, the results clearly indicate that several corrole molecules associate with one protein and that redistribution takes place at higher [HSA]:[corrole] ratios in favor of the 1:1 conjugate. As the number of binding sites in protein-chromophore conjugates and the corresponding binding constants are frequently determined from Scatchard plots (48), we also attempted to do so. However, this method and the underlying equations require the presence of unbound chromophore, while our later indications show that there is no free corrole at the lowest concentrations that can be measured by UV-vis (10-6 M). Accordingly, none of the binding constants can be determined by UV-vis titrations. Effect of HSA on Corrole Fluorescence. Binding of corroles to HSA can be quantified by monitoring the highly intense fluorescence of 2 and 2-Ga, allowing experiments to be performed at significantly lower corrole concentrations. Another major advantage of fluorescence monitoring is that free and bound corroles might have different emission maxima and/or intensities. The differences in emission wavelengths of both 2 and 2-Ga in the presence and absence of HSA are only a few nm, but the fluorescence intensities are very sensitive to the [HSA]:[corrole] ratio (Figure 3). These effects are further demonstrated for selected wavelengths in Figure 4: the initial fluorescence of 2 is reduced to less than 40% at a [HSA]:[2] ratio of 0.1:1 and it is 60% of the initial intensity at a 1:1 ratio. The results for 2-Ga were
Amphiphilic Corroles
Figure 3. Emission spectra of corrole 2 (a, λexc ) 436 nm) and 2-Ga (b, λexc ) 426 nm) as a function of added HSA.
Figure 4. Fluorescence intensity at selected wavelengths of 2 (a) and 2-Ga (b) as a function of the HSA:corrole ratio.
even more pronounced: about 40% reduction at a [HSA]:[2-Ga] ratio of 0.1:1 and 130% of the initial intensity at a 1.1:1 ratio. The latter phenomenon suggests that the extent of nonradiative decay of the excited state is lowered upon binding of the fluorophore to the protein (less degrees of freedom), reflected in increased fluorescence. Nevertheless, the intensity becomes lower when many corrole molecules are bound to the same protein because of self-quenching. For quantitative elucidation of the dissociation constants (Kd) assigned to the high-affinity binding sites, the fluorescence intensities of equimolar conjugates were measured as a function of concentration, with the expectation of a nonlinear correlation if there is significant dissociation. However, a linear correlation was obtained in the range [2-Ga] ) [HSA] ) 2 × 10-6 to 4 × 10-8 M, thereby confirming that no significant dissociation takes place, even at very low (40 nM) conjugate concentrations. Circular Dichroism (CD). CD spectra of corrole solutions as a function of HSA concentration are shown
Bioconjugate Chem., Vol. 15, No. 4, 2004 741
in Figure 5. Since HSA does not absorb visible light, the CD spectrum at long wavelengths (λ > 360 nm) can only be attributed to corrole interactions with the chiral environment of the folded protein. Intriguingly, the 2-Mn CD signal (Figure 5c) is centered at around 490 nm, which is the region of the spectrum that is most sensitive to axial ligands. Moreover, the response to HSA addition is irregular, as can be seen from the shifts of λmax and the inversion from negative to positive ellipticity at 482 nm. This phenomenon is amplified at 440 and 425 nm, respectively, in the spectra of 2 and 2-Ga (Figure 5a,b). These data accord with the earlier conclusions about weak and strong binding sites, which in the present case is observed by redistribution of weakly bound corroles (under the conditions of [corrole] . [HSA]) as the [corrole]:[HSA] ratio approaches 1:1. The results show that there are multiple chiral binding sites, albeit with significant differences in wavelengths and ellipticities. CD spectra of HSA in the region 200-250 nm as a function of added 2-Ga also were obtained to monitor possible changes of protein secondary structure in response to corrole binding (49). As can be seen in Figure 6, HSA gives rise to a strong CD signal at 222 nm, attributable to R-helical structures. The intensity of the signal drops as a function of the [2-Ga]:[HSA] ratio (0.4%, 1.1%, 1.5%, 2% at 1:1, 2:1, 3:1, and 4:1, respectively), and, at a ratio of 10:1, roughly 8% of helical structure is lost. Although there is some degree of conformational change, these corrole-induced perturbations of protein secondary structure are smaller than those observed for BSAporphyrin conjugates (48). HPLC of HSA:Corrole Conjugates. There are many precedents in which serum albumins are used to affect the retention time of ligands with which they interact strongly, including even the immobilization of BSA or HSA on HPLC stationary phases (50). However, we are not aware of any report about noncovalent conjugates of HSA remaining intact during HPLC separation. Apparently, under the high dilution conditions intrinsic to the experiment, most conjugates dissociate. Nevertheless, because of the aforementioned indications of extremely low dissociation constants of the noncovalent HSA/corrole conjugates, we decided to examine them by HPLC. Separate injections revealed that HSA is eluted much faster (1 min, indicative of negligible interaction with the solid phase) than any of the corroles (9-10 min) on the reversed-phase column. Next, HPLC chromatograms of premixed mixtures of HSA and the corroles were examined, setting the detector wavelength at λ > 400 nm (436, and 426, and 486 nm for 2, 2-Ga, and 2-Mn, respectively) as to determine if the eluted HSA fractions contain bound corrole. The results shown in Figure 7 clearly demonstrate the following phenomena: the fastest eluting peak (at 1-2 min) corresponds to HSA-bound corrole, while the two later peaks (at ∼8 min and ∼9 min) to corrole that dissociated from HSA during chromatography (50). For an initial ratio of [HSA]:[corrole] of 5:1, the percentage of corrole that remained bound to HSA was 80, 70, and 40% for 2, 2-Ga, and 2-Mn, respectively. (With 0.1 M phosphate buffer solutions of pH 7.0 instead of unbuffered water, the percentage of corrole that remained bound to HSA was 74, 66, and 46% for 2, 2-Ga, and 2-Mn, respectively). This indicates that the relative dissociation constants (Kd) of the conjugates are in the order 2 e 2-Ga < 2-Mn. Since an excess of HSA was used, we may safely assume that this trend reflects the Kd of particularly strong binding sites. This assumption was checked by the following experiments: injection of mixtures that contain an initial 1:5 ratio of
742 Bioconjugate Chem., Vol. 15, No. 4, 2004
Mahammed et al.
Figure 5. HSA titrations of (a) 2, (b) 2-Ga, and (c) 2-Mn, followed by CD spectroscopy. [2] ) 2.11 × 10-5 M, [2-Mn] ) 10-4 M, [2-Ga] ) 2.7 × 10-5 M; [HSA] ) 0-10-4 M.
Figure 6. (a, b) CD spectra of HSA as a function of [2-Ga]: [HSA] (0, 1, 2...10) at constant [HSA] (3.7 × 10-6 M).
[HSA]:[corrole], isolation of the HSA containing fraction, and examination of the [HSA]:[corrole] therein by recording the full electronic spectrum and relying on
the corresponding values. The results show that [HSA]:[corrole] ratios in the HPLC-eluted conjugates were 1:0.5, 1:0.4, and 1:0.2 for 2, 2-Ga, and 2-Mn, respectively. Apparently, the final [HSA]:[corrole] ratio did not exceed 1:1 even with corrole in excess; thus, the Kd trend 2 e 2-Ga < 2-Mn reflects only the high-affinity binding sites. Since hemin also has been reported to form a tight conjugate with HSA (Kd ) 10-8 M), it also was analyzed by HPLC. At a [HSA]:[hemin] ratio of 5:1, 92% of hemin remained HSA-bound, and when the initial ratio of [HSA]:[hemin] was 1:5, the [HSA]:[hemin] ratio in the HPLC-eluted conjugates was 1: 0.5. All these results show that HPLC experiments provide a good tool for estimating noncovalent binding of porphyrinoids to proteins. Nevertheless, the acquired information must be considered as qualitative rather than quantitative, since other factors such as the water solubility of different porphyrinoids (hemin vs corrole in the current case) may affect the results significantly. Quenching of Trp214 Emission. One of the earliest and simplest methods that was used for investigating hemin binding is fluorescence quenching of the single tryptophan residue (Trp214) in HSA (40). The advantage of this approach is that it reports selectively on binding sites that are not too distant from the fluorophore. Results obtained by exciting HSA’s Trp214 at 280 nm and recording its emission (λmax ) 340 nm) as a function of the concentration of 2-Ga are shown in Figure 8a. Notably, full and selective quenching was achieved, as can be seen from the residual emission (λmax ) 310 nm) due to the many tyrosine residues that are present in
Amphiphilic Corroles
Bioconjugate Chem., Vol. 15, No. 4, 2004 743
Figure 8. Quenching of the emission of HSA by (a) increasing amounts of 2-Ga and (b) the dependence of the emission on the [corrole]/[HSA] ratio. λexc ) 280 nm, [HSA] ) 3.3 × 10-6 M. Figure 7. HPLC chromatograms of HSA conjugates (a) 2, (b) 2-Ga, and (c) 2-Mn, ([HSA]0 ) 2.5 × 10-4 M, [corrole]0 ) 5 × 10-5 M) with the detector set at the wavelength where the absorbance of free and conjugated corroles are identical (436, 426, 486 nm for a, b, c).
HSA. More quantitative information was obtained from plots of fluorescence intensity as a function of the [HSA]: [corrole] ratio (Figure 8b). The responses (slopes) of 2 and 2-Ga are practically identical, and that of 2-Mn somewhat smaller. The up to 50% quenching at equimolar concentrations of HSA and corrole obtained in the first two cases implies that 2 and 2-Ga occupy close to Trp214 binding sites with very high-affinity. If 2-Mn also associates to similar binding sites, Figure 8b may be analyzed as reflecting an opposite relationship between the steepness of the slopes and the Kd of the HSA:corrole conjugates. This would lead to relative Kd from highaffinity sites in the order of 2 ∼ 2-Ga < 2-Mn, in line with the previously mentioned indications. The roughly 50% quenching at a 1:1 ratio between HSA and 2 or 2-Ga may be interpreted either one of two limiting ways: 100% quenching efficiency by sites that are only 50% occupied () large Kd) or 50% quenching efficiency by sites that are 100% occupied () small Kd). Two sets of experiments were performed to distinguish between these possibilities. The first consisted of dilution experiments on 1:1 mixtures of HSA and the corroles. For all three corroles (data for 2-Ga are shown in the inset of Figure 9) there was no deviation from linearity in the concentration vs fluorescence-intensity plots in the range 4 × 10-6 to 7 × 10-7 M (Kd < 10-8 M for the conjugates). Figure 9 shows the other set of experiments: measuring fluorescence intensity of HSA as a function of the relative amount of 2-Ga, at four different absolute concentrations. The normalized plots are identical as long as the corrole is not in excess of HSA, clearly confirming the absence of any dissociation. The plots become different only after
Figure 9. Normalized fluorescence intensity of HSA as a function of the [2-Ga]:[HSA] ratio at different initial concentrations of HSA. Inset: Fluorescence intensity at 340 nm of equimolar concentrations of 2-Ga and HSA at different absolute concentrations.
the [2-Ga]:[HSA] is above 1, indicating the presence of slightly weaker secondary binding sites. Importantly, although a reliable measurement of Kd is problematic because of the large number of corrole molecules that can be accommodated by HSA, the trends of Figure 9 show that dissociation from weaker binding sites is very small as well. Therefore, the approximately 50% quenching seen at a 1:1 ratio between HSA and corrole is consistent with incomplete quenching of Trp214 fluorescence by corroles that occupy high affinity sites.
744 Bioconjugate Chem., Vol. 15, No. 4, 2004
Mahammed et al.
Figure 10. (a) Structural models depicting four potential binding sites for 2-Ga to HSA. (b) 2-Ga in place of the heme of methemalbumin. (c) 2-Ga coordinated to His3. (d) 2-Ga coordinated to His105. (e) 2-Ga coordinated to His146. The proposed locations were chosen as follows: one is the heme binding site; others are histidine binding sites whose positions are consistent with the Trp214 quenching experiments, except for e.
Binding Sites for 2-Ga. Since the three-dimensional structure of HSA is known, we have attempted to
estimate the distance between Trp214 and a corrole binding site via standard Fo¨rster energy transfer equa-
Amphiphilic Corroles
Bioconjugate Chem., Vol. 15, No. 4, 2004 745
tions, using the roughly 50% fluorescence quenching by 2-Ga.
Ro6 ) 8.8 × 10-5(κ2n-4φoJ)
(1)
∫0 Fo(λ)EA(λ) λ4dλ J) ∫0∞Fo(λ)dλ
(2)
∞
Equations 1 and 2 were used for evaluation of the Fo¨rster distance (51, 52), where J is the overlap between the luminescence spectrum of the donor and the absorption spectrum of the acceptor (weighted by λ4), φo is the luminescence quantum yield in the absence of energy transfer, n is the refractive index, and κ is an orientation factor dependent on the alignment of the donor and acceptor dipoles (κ2 ) 2/3 for random alignment). According to the above expressions, our experimental data yielded a value for the overlap integral J between the fluorescence spectrum of the protein HSA and the absorption spectrum of the quencher 2-Ga of 6.26 × 1014. Based on the J value and eq 1, the Fo¨rster distance (Ro) between 2-Ga and Trp214 (HSA) was calculated to be 30 Å. Moreover, models of the 2-Ga-HSA conjugate based on crystallographic data for [Ga(tpfc)(py)] (tpfc ) 5,10,15-tris(pentafluorophenyl)corrolate; py ) pyridine) and HSA were constructed (53). In the HSA crystal structure, a heme (FeIII-protoporphyrin IX) is located in a hydrophobic pocket in subdomain IB, and there are additional interactions involving Tyr161, Ile142, Tyr138, His146, and Lys190 with the heme. In one of our proposed structures, where noncovalent interactions are dominant (Figure 10a), 2-Ga replaces the heme in the hydrophobic cavity; at this site, the distances between 2-Ga and Trp214 are 20-24 Å (from the edge of the corrole-from the metal), in fair agreement with the estimated distance based on fluorescence quenching. Figures 10b and 10c show two 2-Ga-HSA interactions that involve metalcoordination to His. As examples, we propose that 2-Ga could bind to surface-exposed His3 and His105 with distances between His-bound 2-Ga and Trp214 of 24-30 and 30-34 Å, respectively. We also note that coordination of 2-Ga to His146 located in the heme binding pocket (Figure 10d) is unlikely due to a considerable difference between the Fo¨rster distances measured in the model and that estimated from quenching experiments (16-19 Å vs 30 Å). CONCLUDING REMARKS
We report results from an investigation of interactions between corroles and biomolecules. Corrole 2 is especially promising for biomedical research because of its structural similarity to the most active porphyrin-based drugs and because amphiphilicity is easily achieved. Our work has focused on interactions with HSA because any medical application must take into account associations with the most abundant protein in serum. Two metal complexes of 2 were investigated, with the goal of elucidating any effects due to ligation to relevant residues of the protein. The changes in the electronic absorption spectra of 2, 2-Ga, and 2-Mn as well as in the emission spectra of 2 and 2-Ga as a function of increasing amounts of HSA revealed many protein binding sites. With corrole in excess, both absorption and emission intensities decreased, which is behavior that is frequently encountered when porphyrins form aggregates, usually because of limited solubility in water. However, the corroles of the
current investigation do not aggregate in aqueous solutions, and, accordingly, we propose that the above phenomena are due to self-quenching of many corrole molecules that are aggregated/associated in binding pockets within the protein. Examination of published crystal structures of HSA reveals the presence of a large calixarene-like amphiphilic pocket that seems appropriate for accommodation of the corroles. When the [corrole]: [HSA] ratio approaches 1:1, the absorption and emission intensities increase, thereby indicating redistribution of corroles from the weaker binding pocket to high-affinity sites. We have used many different methods to investigate the interactions of corroles with HSA. The results clearly show that all derivatives form tightly bound conjugates with the protein, with dissociation constants in the nanomolar range in the order 2 e 2-Ga < 2-Mn. Under biologically relevant conditions (HSA concentrations 2 to 3 orders of magnitude larger than those examined here), all the derivatives will be fully bound to proteins. This information is of prime importance for any potential utilization of the corroles in bioinorganic systems such as for therapeutic applications and biomimetic catalysis, approaches that are currently under investigation in our laboratories. ACKNOWLEDGMENT
Work at the Technion was supported by the STAR (Chicago) Foundation (Z.G.) and the Israel Science Foundation (Z.G.); work at Caltech was supported by JPLNASA (K.S.) and the National Science Foundation (H.B.G., J.J.W.). LITERATURE CITED (1) Sessler, J. L., Tvermoes, N. A., Davis, J., Anzenbacher, P., Jursikova, K., Sato, W., Seidel, D., Lynch, V., Black, C. B., Try, A., Andrioletti, B., Hemmi, G., Mody, T. D., Magda, D. J., and Kral, V. (1999) Expanded porphyrins. Synthetic materials with potential medical utility. Pure Appl. Chem. 71, 2009-2018. (2) MacDonald, I. J. and Dougherty, T. J. (2001) Basic principles of photodynamic therapy. J. Porphyrins Phthalocyanines 5, 105-129. (3) Via, L. D., and Magno, S. M. (2001) Photochemotherapy in the treatment of cancer. Curr. Med. Chem. 8, 1405-1418. (4) Dougherty, T. J., Gomer, C. J., Henderson, B. W., Jori, G., Kessel, D., Korbelik, M., Moan, J., and Peng, Q. (1998) Photodynamic therapy. J. Natl. Cancer. Inst. 90, 889-905. (5) Sessler, J. L., and Weghorn, S. J. (1997) Expanded, Contracted, & Isomeric Porphyrins, pp 1-9, Pergamon, Oxford. (6) Paolesse, R. (2000) Syntheses of Corroles. The Porphyrin Handbook, Vol. II. (Kadish, K. M., Smith, K. M., and Guilard, R., Eds.) pp 201-232, Academic Press, New York. (7) Gross, Z., Galili, N., and Saltsman, I. (1999) The first direct synthesis of corroles from pyrrole. Angew. Chem., Int. Ed. 38, 1427-1429. (8) Gross, Z., Galili, N., Simkhovich, L., Saltsman, I., Botoshansky, M., Bla¨ser, D., Boese, R., and Goldberg, I. (1999) Solventfree condensation of pyrrole and pentafluorobenzaldehyde: A novel synthetic pathway to corrole and oligopyrromethenes. Org. Lett. 1, 599-602. (9) Paolesse, R., Jaquinod, L., Nurco, D. J., Mini, S., Sagone, F., Boschi, T., and Smith, K. M. (1999) 5,10,15-triphenylcorrole: a product from a modified rothemund reaction. Chem. Commun. 1307-1308. (10) Simkhovich, L., Goldberg, I., and Gross, Z. (2000) First syntheses and X-ray structures of a meso-alkyl substituted corrole and its Ga(III) complex. J. Inorg. Biochem. 80, 235238. (11) Cho, W.-S., and Lee, C.-H. (2000) Customized, coremodified corroles from [2+2] condensation of dipyrromethanes. Tetrahedron Lett. 41, 697-701.
746 Bioconjugate Chem., Vol. 15, No. 4, 2004 (12) Gryko, D. T. (2000) A simple, rational synthesis of mesosubstituted A(2)B-corroles. Chem. Commun. 2243-2244. (13) Brin˜as, R. P., and Bru¨ckner, C. (2001) Triarylcorroles by oxidative coupling of triaryltetrapyrranes. Synlett 442-444. (14) Gryko, D. T., and Jadach, K. (2001) A simple and versatile one-pot synthesis of meso-substituted trans-A(2)B-corroles. J. Org. Chem. 66, 4267-4275. (15) Ka, J.-W., Cho, W.-S., and Lee, C.-H. (2000) Expedient synthesis of corroles by oxidant-mediated, direct R-R′ coupling of tetrapyrromethanes. Tetrahedron Lett. 41, 81218125. (16) Paolesse, R., Nardis, S., Sagone, F., and Khoury, R. G. (2001) Synthesis and functionalization of meso-aryl-substituted corroles. J. Org. Chem. 66, 550-556. (17) Asokan, C. V., Smeets, S., and Dehaen, W. (2001) Sterically encumbered triarylcorroles from aryldipyrromethanes and aromatic aldehydes. Tetrahedron Lett. 42, 4483-4485. (18) Gross, Z., Simkhovich, L., and Galili, N. (1999) First Catalysis by corrole metal complexes: epoxidation, hydroxylation, and cyclopropanation. J. Chem. Soc., Chem. Commun. 599-600. (19) Gross, Z., Golubkov, G., and Simkhovich, L. (2000) Epoxidation catalysis by a manganese(III) corrole and isolation of an oxomanganese(V) corrole. Angew. Chem., Int. Ed. 39, 4045-4047. (20) Simkhovich, L., Mahammed, A., Goldberg, I., and Gross, Z. (2001) Synthesis and characterization of germanium, tin, phosphorous, iron, and rhodium complexes of tris(pentafluorophenyl)corrole, and the utilization of the iron and rhodium corroles as cyclopropanation catalysts. Chem. Eur. J. 7, 10411055. (21) Golubkov, G., Bendix, J., Gray, H. B., Mahammed, A., Goldberg, I., Di Bilio, A. J., and Gross, Z. (2001) High valent manganese corroles and the first perhalogenated corrolebased catalyst. Angew. Chem., Int. Ed. 40, 2132-2134. (22) Simkhovich, L., and Gross, Z. (2001) Iron(IV) corroles are potent catalysts for aziridination of olefins by chloramine-T. Tetrahedron Lett. 42, 8089-8092. (23) Bendix, J., Dmochowski, I. J., Gray, H. B., Mahammed, A., Simkhovich, L., and Gross, Z. (2000) Structural, electrochemical, and photophysical properties of gallium(III) 5,10,15-tris(pentafluorophenyl)corrole. Angew. Chem., Int. Ed. 39, 4048-4051. (24) Mahammed, A., and Gross, Z. (2002) Aluminum corrolin, a novel chlorophyll analogue. J. Inorg. Biochem. 88, 305-309. (25) Mahammed, A., Goldberg, I., and Gross, Z. (2001) Highly selective chlorosulfonation of tris(pentafluorophenyl)corrole as a synthetic tool for the preparation of amphiphilic corroles and metal complexes of chiral planarity. Org. Lett. 3, 34433446. (26) Saltsman, I., Mahammed, A., Goldberg, I., Tkachenko, E., Botoshansky, M., and Gross, Z. (2002) Selective substitution of corroles: nitration, hydroformylation, and chlorosulfonation. J. Am. Chem. Soc. 124, 7411-7420. (27) Allen, C. M., Langlois, R., Sharman, W. M., La Madeleine, C., and Van Lier, J. E. (2002) Photodynamic properties of amphiphilic derivatives of aluminum tetrasulfophthalocyanine. Photochem. Photobiol. 76, 208-216. (28) Boyle, R. W., and Dolphin, D. (1996) Structure and biodistribution relationships of photodynamic sensitizers. Photochem. Photobiol. 64, 469-485. (29) Bonnett, R. (1999) Photodynamic therapy in historical perspective. Rev. Contemp. Pharmacother. 10, 1-17. (30) Chuang, V. T. G., Kragh-Hansen, U., and Otagiri, M. (2002) Pharmaceutical strategies utilizing recombinant human serum albumin. Pharm. Res. 19, 569-577. (31) Kragh-Hansen, U., Chuang, V. T. G., and Otagiri, M. (2002) Practical aspects of the ligand-binding and enzymatic properties of human serum albumin. Biol. Pharm. Bull. 25, 695704. (32) Peters, T. (1996) All About Albumin: Biochemistry, Genetics, and Medical Applications, Academic Press, San Diego. (33) Sugio, S., Kashima, A., Mochizuki, S., Noda, M., and Kobayashi, K. (1999) Crystal structure of human serum albumin at 2.5 angstrom resolution. Protein Eng. 12, 439446 and references therein.
Mahammed et al. (34) Wardell, M., Wang, Z., Ho, J. X., Robert, J., Ruker, F., Ruble, J., and Carter, D. C. (2002) The atomic structure of human methemalbumin at 1.9 angstrom. Biochem. Biophys. Res. Commun. 291, 813-819. (35) Bertucci, C. and Domenici, E. (2002) Reversible and covalent binding of drugs to human serum albumin: Methodological approaches and physiological relevance. Curr. Med. Chem. 9, 1463-1481. (36) Ding, Y. S., Lin, B. C., and Huie, C. W. (2001) Binding sudies of porphyrins to human serum albumin using affinity capillary electrophoresis. Electrophoresis 22, 2210-2216. (37) Morgan, W. T., Smith, A., and Koskelo, P. (1980) The interaction of human serum albumin and hemopexin with porphyrins Biochim. Biophys. Acta 624, 271-285. (38) Datta-Gupta, N., Malakar, D., and Dozier, J. (1989) Binding studies of 4 free base porphyrins and 6 Fe(III) porphyrins with human serum albumin. Res. Commun. Chem. Pathol. Pharmacol. 63, 289-292. (39) Rotenberg, M., Cohen, S., and Margalit, R. (1987) Thermodynamics of porphyrin binding to serum-albumin - Effects of temperature, of porphyrin species and of albumin-carried fatty-acids. Photochem. Photobiol. 46, 689-693. (40) Beaven, G. H., Chen, S. H., Dalbis, A., and Garatzer, B. W. (1974) Spectroscopic study of hemin-human serum albumin system. Eur. J. Biochem. 41, 539-546. (41) Bardos-Nagy, I., Galantai, R., Kaposi, A. D., and Fidy, J. (1998) Difference in the transport of metal and free-base porphyrins: Steady-state and time-resolved fluorescence studies. Int. J. Pharm. 175, 255-267. (42) Gantchev, T. G., Ouellet, R., and van Lier, J. E. (1999) Binding interactions and conformational changes induced by sulfonated aluminum phthalocyanines in human serum albumin. Arch. Biochem. Biophys. 366, 21-30. (43) Ambroz, M., MacRobert, A. J., Morgan, J., Rumbles, G., Foley, M. S. C., and Phillips, D. (1994) Time-resolved fluorescence spectroscopy and intracellular imaging of disulfonated aluminum phthalocyanine. J. Photochem. Photobiol. B22, 105-117. (44) Andrade, S. M., and Costa, S. M. B. (2002) Spectroscopic studies on the interaction of a water soluble porphyrin and two drug carrier proteins. Biophys. J. 82, 1607-1619. (45) Davila, J., and Harriman, A. (1990) Photochemical and radiolytic oxidation of a zinc porphyrin bound to human serum albumin. J. Am. Chem. Soc. 112, 2680-2690. (46) Bishop, S. M., Khoo, B. J., MacRobert, A. J., Simpson, M. S. C., and Phillips, D. (1993) Characterization of the photochemotherapeutic agent disulfonated aluminum phthalocyanine and its high-performance liquid-chromatographic separated components. J. Chromatogr. 646, 345-350. (47) Mahammed, A., Gray, H. B., Weaver, J., and Gross, Z. (2003) How acidic are corroles and why? Tetrahedron Lett. 44, 2077-2079. (48) Chatterjee, S., and Srivastava, T. S. (2000) Spectral investigations of the interaction of some porphyrins with bovine serum albumin. J. Porphyrins Phthalocyanines 4, 147-157. (49) Cai, X. and Dass, C. (2003) Conformational analysis of proteins and peptides. Curr. Org. Chem. 7, 1841-1854. (50) Fitos, I., Visy, J., and Simonyi, M. J. (2002) Speciesdependency in chiral-drug recognition of serum albumin studied by chromatographic methods. J. Biochem. Biophys. Methods 54, 71-84 and references therein. (51) Dunn, A. R., Dmochowski, I. J., Winkler, J. R., and Gray, H. B. (2003) Nanosecond photoreduction of cytochrome P450cam by channel-specific Ru-diimine electron tunneling wires. J. Am. Chem. Soc. 125, 12450-12456. (52) Turro, N. J. (1978) Modern Molecular Photochemistry, Benjamin/Cummings Inc., Menlo Park. (53) Structural models were built using the Accelrys DS ViewerPro 5.0 software package. The structure of 2-Ga was obtained by addition of two sulfonate groups and removal of the axial pyridine from the published crystal structure (ref 23): 2-Ga was then inserted into putative HSA binding sites (crystal structure from ref 34). BC034179P
